Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 55 of the 55 references it cites were found in a registry or an index; none of the 5 sources its claims cite is checked for support yet.

ATEM: A Topic Evolution Model for the Detection of Emerging Topics in Scientific Archives

Open access
Rahimi, Hamed, Naacke, Hubert, Constantin, Camelia, Amann, Bernd · 2024 · Studies in Computational Intelligence

Overview

CoChat Summary

TL;DR

ATEM is a framework for detecting emerging interdisciplinary topics in scientific archives by combining dynamic topic modeling with dynamic graph embedding over topic-citation graphs. Experiments on DBLP (over five million computer science articles, 2000–2020) show that citation-context embedding neighbors yield higher emergence predictability than co-citation-linked topic pairs.

Key findings

  • ATEM can efficiently detect emerging cross-disciplinary topics within the DBLP archive of over five million computer science articles.
  • In a proof-of-concept sample of 200 evolving topics with n=10 topics per period, NeighborTopics (new embedding nearest neighbors, max distance 0.2, min embedding norm 0.22) had higher predictability than LinkedTopics (citation-path-connected pairs). On average, 75% of emerging topics have 1.25/0.75 = 1.66 times more publications after emergence than before; 50% of NeighborTopics have 2.6/0.4 = 6.2 more publications after emergence than before, whereas the ratio is 1 for LinkedTopics.
  • The average number of NeighborTopics decreases over time and average embedding distance increases; predictability decreases with increasing distance and strongly decreases in time; average emergence is positive before 2016 and becomes negative afterward.
  • Example: topics T680C6 and T661C6 emerge in 2013 as near embedding neighbors with a maximal distance of 0.2, and most common documents for the pair were published after 2013.

Methods

  • ATEM extracts evolving topics by combining content-based document clustering (Doc2Vec embeddings, UMAP, HDBSCAN) and citation-based clustering (Leiden community detection), intersecting the two cluster types, splitting clusters into time periods, and representing topics via nearest words and class-based TF-IDF.
  • It builds a dynamic topic-citation graph by projecting document citation links into topic space, then applies OnlineNode2Vec dynamic node embeddings and cosine similarity to define emerging topics as topic pairs whose context distance crosses a threshold at a given period.
  • The implementation is applied to DBLP with 5M scientific articles published between 2000 and 2020; the proof of concept randomly samples 200 evolving topics, generates NeighborTopics and LinkedTopics sets of n=10 per period, and compares their predictability based on document distribution before and after emergence.

Limitations

  • Stated: the conclusion that analyzed topics get more diverse over time must be confirmed by deeper analysis of bias introduced by the dynamic computation algorithm; emergence values are influenced by the relative length of past and future covered by the archive, with average emergence positive before 2016 and negative afterward; future work is needed on other datasets and domains.
  • Evident: the evaluation is a proof of concept on one dataset with one main comparison baseline (co-citation/LinkedTopics), and no ground-truth emerging topics or precision/recall metrics are reported; the predictability metric may be affected by archive time coverage and embedding thresholds.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1007/978-3-031-53472-0_28

Cites

55 of 55 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 The Architecture of ATEM. This framework extracts evolving topics from a corpus of documents using a dynamic topic model. It then creates a dynamic topic-citation graph by projecting document citation links into the topic space. Finally, it applies a dynamic graph embedding method on the topic-citation graph for representing the dynamic citation context of topics. Based on this context, emerging topics can be defined as couples or sets of evolving topics with similar citation contexts.
Figure 1 The Architecture of ATEM. This framework extracts evolving topics from a corpus of documents using a dynamic topic model. It then creates a dynamic topic-citation graph by projecting document citation links into the topic space. Finally, it applies a dynamic graph embedding method on the topic-citation graph for representing the dynamic citation context of topics. Based on this context, emerging topics can be defined as couples or sets of evolving topics with similar citation contexts.CC-BY · From the PDF
Figure 2 Evolution of citations of evolving topic T680C6.
Figure 2 Evolution of citations of evolving topic T680C6.CC-BY · From the PDF
Figure 3 Evolution of citations of evolving topic T485C6
Figure 3 Evolution of citations of evolving topic T485C6CC-BY · From the PDF
Figure 4 Co-citations of T680C6 and T485C6
Figure 4 Co-citations of T680C6 and T485C6CC-BY · From the PDF
Figure 5 Citing evolving topic T70C6
Figure 5 Citing evolving topic T70C6CC-BY · From the PDF
Figure 6 Embedding distance evolution for topic T680C6 at period 2013
Figure 6 Embedding distance evolution for topic T680C6 at period 2013CC-BY · From the PDF
Figure 7 Evolving topic T661C6
Figure 7 Evolving topic T661C6CC-BY · From the PDF
Figure 8 The Implementation of Extracting Evolving Topics.
Figure 8 The Implementation of Extracting Evolving Topics.CC-BY · From the PDF
Image 9
CC-BY · From the PDF
Image 10
CC-BY · From the PDF
Figure 11 Violin distribution of predictability values.
Figure 11 Violin distribution of predictability values.CC-BY · From the PDF
Figure 12 The distance distribution of emerging topics.
Figure 12 The distance distribution of emerging topics.CC-BY · From the PDF
Figure 13 The average of predictability values by year
Figure 13 The average of predictability values by yearCC-BY · From the PDF
Figure 14 The correlation between emerging topic properties
Figure 14 The correlation between emerging topic propertiesCC-BY · From the PDF
Figure 15 The correlations between 𝑁𝑒𝑖𝑔ℎ𝑏𝑜𝑟𝑇𝑜𝑝𝑖𝑐𝑠 , 𝐿𝑖𝑛𝑘𝑒𝑑𝑇𝑜𝑝𝑖𝑐𝑠 and their intersection.
Figure 15 The correlations between 𝑁𝑒𝑖𝑔ℎ𝑏𝑜𝑟𝑇𝑜𝑝𝑖𝑐𝑠 , 𝐿𝑖𝑛𝑘𝑒𝑑𝑇𝑜𝑝𝑖𝑐𝑠 and their intersection.CC-BY · From the PDF

Verification

Claims the paper makes

0 Valid · 0 Not valid · 3 Not checked · 36 Own results and interpretations

20 more claims

Create a free account to see all 39 claims and to verify the 5 references still open.

See all claims — free account
  • The quote only partly states this claim. 1
  • It may state two separate claims as one. 4
  • Probably not a substantial claim. 10

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 51 references

Cited by

See all 9

Metadata

Authors
Rahimi, H., Naacke, H., Constantin, C., & Amann, B.
Affiliations
Centre National de la Recherche Scientifique (FR)
Corresponding author
Rahimi, Hamed
Venue
Studies in Computational Intelligence
Published
2024
Volume
1143
Pages
332-343
ISSN
  • 1860-949X (ISSN-L)
  • 1860-9503
OpenAlex
W4391973680
Cited by
9
References
51
Citations by year
2026: 3 · 2025: 3 · 2024: 3
FWCI
0.78
Type
conference paper
Language
en
Open access
Licence
Locations
  • Studies in computational intelligence (publishedVersion)
  • HAL (submittedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • HAL (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
  • Social Sciences › Library and Information Sciences › scientometrics and bibliometrics research
  • Social Sciences › General Social Sciences › Computational and Text Analysis Methods
Keywords
topic evolution, emerging research areas, dynamic topic modeling, dynamic graph embedding, citation context, citation dynamics, DBLP, computer science literature, scientific archives

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-18, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-18 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: ATEM: A Topic Evolution Model for the Detection of Emerging Topics in Scientific Archives. https://app.cochat.ai/papers/10.1007/978-3-031-53472-0_28
The paper
Rahimi, H., Naacke, H., Constantin, C., & Amann, B. (2024). ATEM: A Topic Evolution Model for the Detection of Emerging Topics in Scientific Archives. Studies in Computational Intelligence. https://doi.org/10.1007/978-3-031-53472-0_28

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.